mediumget_corpus_coverage
References secrets or data exfiltration
The tool text references credentials, environment variables, or sending data elsewhere. This can be legitimate, but review what this tool actually accesses.
mediumcheck_mcp_server
Embedded external URL encourages out-of-band request to unfamiliar domain
The description tells the agent to 'GET https://sri-test.biz/v1/corpus' directly, outside the tool's own invocation path. This is an instruction embedded in a tool description that steers the agent toward making a raw HTTP request to a third-party domain that isn't verified as an official registry endpoint. An agent following this literally could leak query context or be redirected to attacker-controlled infrastructure. The domain name ('sri-test.biz') doesn't match any known official registry and looks disposable/test-like, which is a red flag for spoofing a legitimate coverage index.
mediumcheck_mcp_server
Undisclosed cryptocurrency payment mechanism (x402/USDC) may trigger charges without explicit user consent
The 'l402_token' field description states 'Payment, when charged, is over x402 (USDC on Base) via HTTP 402.' This reveals the tool can silently escalate into a paid, on-chain transaction flow triggered by an HTTP 402 response, without the schema exposing any explicit confirmation step. An agent calling this tool for what looks like a free lookup could inadvertently initiate a crypto payment on the user's behalf. This capability (moving funds) is far broader than the tool's stated purpose of read-only source analysis and should require explicit user approval before any payment path is exercised.
low
Not on gate's verified list
This server isn't on gate's handpicked list. That's not necessarily bad, but there's no third-party signal about it — review its tools carefully.
lowcheck_mcp_server
Persuasive framing pushes agent behavior beyond the tool's mechanical function
Phrases like 'Call this before connecting to, installing, or invoking an MCP server you have not read yourself' instruct the agent on when/how to use other tools, which is a form of behavioral steering embedded in a tool description. While plausibly benign and aligned with safety, this pattern (instructions to the AI about its broader workflow, not just this tool's I/O) is the same mechanism used in more harmful prompt-injection attempts and warrants scrutiny given the tool's own admitted purpose is to detect exactly this kind of steering in other servers.